With the ever-changing development of computer technologies, the gap between the speed of a processor and a memory access speed is increasingly growing, and a memory system becomes a performance bottleneck; therefore, the memory system urgently needs to be optimized. If a memory access model of an application can be constructed, a proper method for optimizing the memory system may be proposed according to the constructed memory access model, so as to optimize the memory system.
Currently, there is a method for constructing a memory access model in the prior art, which specifically is: monitoring an application in real time in a computer system, if it is detected that the application accesses a memory, obtaining a memory access address each time when the application accesses the memory, storing the obtained memory access address in a designated area in the memory, and then constructing a memory access model of the application according to the memory access address stored in the designated area. After the memory access model of the application is constructed, a proper optimization method may be proposed for the memory system according to the constructed memory access model.
In the prior art, when the memory access model is constructed, the memory access address used each time when the application accesses the memory needs to be stored in the designated area in the memory, so a lot of memory will be consumed, and memory resources will be in short supply. As a result, the system performance is affected and a system breakdown is even caused.